Skip to content

Ultimate Security & Trust Boundary Doctrine ​

RULE: Security in OpenSIN is not a checklist item. It is the discipline of deciding what the system is allowed to trust, what it is never allowed to expose, and how it must fail when trust is uncertain.


1. Trust Boundaries Must Be Explicit ​

Every system must know:

  • what input is trusted
  • what input is untrusted
  • what credentials it may access
  • what side effects require confirmation
  • what should fail closed instead of fail open

If these boundaries are implicit, they will be violated.


2. Secret Handling Rules ​

Never do ​

  • hardcode secrets
  • commit secrets
  • print secrets in logs
  • pass secrets through screenshots or issue bodies
  • leave secrets in temporary scripts or throwaway files

Always do ​

  • keep secrets in env vars / credential stores / approved secret surfaces
  • redact logs
  • rotate leaked tokens immediately
  • prefer least privilege

Why ​

A leaked secret is not “just a mistake.” It is an attacker invitation.


3. Confirmation Gates ​

Actions that are destructive, security-sensitive, identity-sensitive, or billing-sensitive require explicit confirmation unless the operator has clearly delegated that authority.

Examples:

  • deleting data
  • rotating production credentials
  • changing auth rules
  • binding domains
  • making billing-affecting API changes

Why ​

Autonomy without permission gates becomes self-sabotage.


4. Fail-Closed by Default ​

When validation is missing, signatures are absent, auth is unclear, or required config is incomplete, the system must refuse activation.

Why ​

Fail-open behavior turns uncertainty into compromise.


5. Scope & Least Privilege ​

Every token, credential, and integration should have only the permissions it actually needs.

Anti-patterns ​

  • one global god-token for everything
  • sharing admin credentials across unrelated agents
  • reusing powerful tokens in docs/tests/examples

6. Security Logging ​

Security-relevant events must be visible without exposing the secret itself:

  • auth failures
  • permission denials
  • suspicious retries
  • unexpected environment mismatch
  • token rotation events

Log the event, not the credential.


7. Browser / UI Security Implications ​

Browser automation often crosses powerful trust boundaries:

  • admin consoles
  • OAuth approvals
  • dashboards with billing power

Therefore profile selection, permissions, and evidence capture are part of security, not just convenience.


8. Documentation Security Rule ​

Docs may explain secret flow, but must not contain real secrets, token-shaped values, or copy-pasteable privileged credentials.

Why ​

A markdown file in Git is forever unless cleaned aggressively.


9. Final Rule ​

Security is the discipline of protecting trust, not the performance of looking careful. If the system cannot explain why it trusts an action, it should not perform it.


Last updated: 2026-04-10
Status: ACTIVE & MANDATORY
Maintainer: sin-zeus


Relevante Mandate ​

MandatPriorityRegel
Antigravity-Only-10.0KEIN gemini-api Provider
Annahmen-Verbot-5.0KEINE Diagnose ohne Beweis
Box.com Storage0.0Alle Logs zu Box.com, nicht lokal

Guides and concepts for the OpenSIN API.